Artificial intelligence has become the driving force behind one of the biggest technology rallies in modern market history. But as AI chip stocks and memory companies surge to record valuations, investors are increasingly asking a critical question: is this a permanent transformation of the semiconductor industry – or another classic tech bubble nearing its peak?
At the center of the rally is soaring demand for high-bandwidth memory (HBM) and next-generation semiconductors used to train large language models (LLMs). Since the launch of ChatGPT in late 2022, the AI race has pushed tech giants and chipmakers into an unprecedented spending cycle.
Yet beneath the excitement, veteran investors warn that the semiconductor industry has historically been one of the most cyclical sectors in global markets – and AI may not eliminate those risks.
The Explosive Rise of the AI Memory Chip Market
The launch of ChatGPT in December 2022 triggered a structural supply shortage in the memory chip market. Companies such as Google, OpenAI, and Anthropic require enormous amounts of computing power and data storage to train advanced AI systems, fueling demand for HBM chips at a historic pace.
That demand has delivered massive gains for leading memory companies in 2026:
- SK Hynix: Up 186% year-to-date
- SanDisk: Up 156% year-to-date
- Micron Technology: Up 141% year-to-date
- Samsung Electronics: Up 114% year-to-date
The rally has also created heavy concentration risk in regional equity markets. In South Korea, Samsung Electronics and SK Hynix now account for more than 50% of the Kospi index, helping push the market to extreme highs throughout 2025 and 2026.
Structural Shift or Another Semiconductor Bubble?
The bullish case for AI semiconductor stocks is built on one core belief: that artificial intelligence has permanently changed the economics of the memory industry.
Historically, semiconductor markets have been highly cyclical, with supply often continuing to expand even after demand slows. But many investors now argue that AI demand is so large and persistent that the traditional memory cycle may no longer apply.
Nomura remains highly optimistic, forecasting SK Hynix could double to 4 million won over the next 12 months, while Samsung Electronics could rise another 20% to 590,000 won.
Not everyone is convinced.
“In the long run it’s a pretty dreadful industry. I suspect that’s still the case every time people make an argument that the memory cycle is gone… just before it all goes horribly wrong.”
– William de Gale, Portfolio Manager at BlueBox Asset Management
For many veteran fund managers, today’s AI-driven enthusiasm resembles previous periods when investors believed technology had permanently removed economic cycles – right before major corrections followed.
Google’s TurboQuant Could Reshape AI Memory Demand
The biggest threat to the AI memory boom may not come from weaker demand, but from better software.
On March 24, Alphabet’s Google introduced TurboQuant, a new compression technology designed to reduce the amount of memory needed to run large language models by six times.
For AI developers, the breakthrough could dramatically improve efficiency and reduce infrastructure costs. For memory manufacturers, however, it raises a serious concern: if AI systems require less physical memory, long-term demand for HBM chips could weaken.
Following the announcement, major memory stocks sold off sharply.
Deutsche Bank warned investors that semiconductor valuations now assume profit margins will remain elevated without companies over-expanding production capacity. The bank also cautioned that heavy momentum positioning has made AI semiconductor stocks increasingly vulnerable to a sudden market shakeout.
Huawei, Nvidia, and the Growing AI Chip Rivalry
At the same time, competition in the global semiconductor industry is intensifying.
Despite ongoing U.S. sanctions, Huawei recently unveiled a new engineering approach called “LogicFolding” for its Kirin smartphone chips. The technology is expected to launch this fall and aims to work around export restrictions that currently limit Nvidia’s ability to sell advanced AI hardware in China.
Meanwhile, TSMC has already entered volume production of true 2-nanometer chips. Huawei claims its stacked “LogicFolding” architecture could eventually deliver capabilities comparable to a 1.4-nanometer process by 2031.
Many analysts remain skeptical of those claims.
Paul Triolo, head of technology for Asia and the Americas at DGA Group, noted that while folded chip architectures can improve density efficiency, they do not solve the underlying manufacturing challenges tied to true 1.4 nm-class production, including power consumption, heat management, yields, and device performance.
Still, Huawei’s semiconductor progress is becoming a growing competitive threat to Apple in China. The company’s Mate 60 smartphone, launched in 2023 with an advanced 5G chip, helped Huawei regain market share from the iPhone. Its upcoming Kirin chip release could intensify that rivalry further.
What Investors Should Know About AI Semiconductor Risks
With software innovation reshaping infrastructure demand and hardware competition accelerating globally, some portfolio managers are becoming increasingly defensive toward AI-related equities.
Andrew Lapping, Chief Investment Officer at Ranmore Fund Management, warns that investors may be paying extreme valuations for an industry that has historically generated only average long-term returns on capital.
“A leopard does not often change its spots.”
Steve Brice, Global Chief Investment Officer at Standard Chartered, also warned that optimism surrounding highly concentrated technology markets – particularly South Korea – may be approaching unsustainable levels.
For more cautious investors, the message is becoming clear: after an extraordinary rally in AI and semiconductor stocks, portfolio diversification and risk management may matter more than chasing momentum.
As history has repeatedly shown, even the strongest technology booms eventually face the realities of competition, efficiency gains, and the traditional forces of supply and demand.